ALUMINA LAKES* BY WILDER D. BANCROFT AND ESTHER COE FARNHAbI
Reinmuth and Gordon' have prepared what they claim to be the aluminum salt of Orange I1 acid by several methods. Since many alumina lakes are adsorption complexes and not definite chemical compounds, it seemed desirable to make a phase rule study of the alumina lakes of alizarin and Orange 11. The Orange I1 was furnished us through the courtesy of the duPont Company. The free acid was prepared by the method of Sisley.' Orange I1 acid is soluble enough in water to be used that way; but aliaarinic acid must be used in alcoholic solution. Powdered alumina from Kahlbaum has been calcined and is quite inert to acids. It took up no aliaarinic acid from alcoholic solution and did not react with Orange I1 in aqueous solution. Sulphuric acid (1-20 N) had no action on it in one week. An alumina catalyst was prepared by heating aluminum nitrate in the usual way. It was only colored a faint pink after standing for a week in contact with a saturated solution of alizarinic acid in alcohol. Alumina was then precipitated from aluminum chloride with a slight excess of ammonia. The precipitate was washed five times by decantation, filtered, and dried partially between filter paper. It was then suspended in sufficient alcohol to form a thick suspension, which was filtered through cheese-cloth to give uniformity. This suspension was evaporated to dryness and the solid heated in an oven at 110' for forty-eight hours. The fine powder required very little grinding before use. The alcohol treatment is useful in dehydrating the gel partly. I t thereby prevents the caking which is generally noticed when water suspensions of alumina are dried. A more adsorbent form of alumina was also prepared by precipitating the gel as described and using the alcoholic suspension without further treatment. Pure sublimed alizarinic acid was dissolved in absolute alcohol. A saturated solution was found on drying to contain about 1.7 mg / cc. The solutions were allowed to stand in stoppered flasks for three weeks a t room temperature in contact with 0.2-0.3 g A1203. About half the supernatant liquid was removed, centrifuged, and analyzed by the oxidation method described by Weiser.s The data are given in Table I and Fig. I , where No. I is with dried alumina, Nos. 2 and 3 with two different samples of suspended alumina. In all three cases the concentrations in the solutions vary continuously and we are therefore dealing with adsorption over the ranges covered by the table.
* This work is done under the rogramme now being carried out at Cornell University and sup orted in part by a grant f o m the Heckscher Foundation for the Advancement of Reaesrcg established by August Heckscher a t Cornell University. 1 Coll. Symp. Mon., 7, 161 (1930). * Bull., (3) 25, 862 (1901). 8 J. Phys. Chem., 31, 1824 (1927).
3128
WILDER D. BANCROFT AND ESTHER COE FARNHAM
TABLE I Alumina and Aliearinic Acid in Alcohol Length of run: three weeks
A1
A
=
B
=
hIilligrams aliearinic acid per gram alumina hlilligrams alizarinic acid per cc solution
Dried Alnmina Bi A1
B,
Alumina Suspension ____ €32
.4 3
n3
0.00
116
200
0.07
202
0.33
139 132
0.08 0.36 0.72
0.50
163
0.77
0.72
148
1.02
A2
36.;
0.10
76.5
0.85
34
0.18
95
I .07
45
0.25
81
1.18
53
0.31
I25
1.43
72
0.52
102.5
1.47
248 268
116
FIG.I Aliznrinic Acid in Alcohol and Alumina
In Run 2 there is practically complete exhaustion of the bath at low concentrations and it might be claimed that this represented an insoluble compound which then adsorbed alizarinic acid. The simplest answer to this is that I 16 mg alizarinic acid per gram alumina corresponds approximately to a formula Alzs(Ct4H504), which is absurd. Since Orange I1 acid is moderately soluble in water, the first run was made with an aqueous solution and a suspension of alumina prepared by precipitating an aluminum chloride solution with ammonia and washing five times by decantation. The concentrations of the stock solutions and the solutions in
3’29
ALUMINA LAKES
final equilibrium were determined by the titanous chloride titration method of Knecht,’ which was found to be quite satisfactory. The data are given in Table 11.
TABLE I1 Alumina and Orange I1 Acid in Water Length of run: three weeks A = Milligrams Orange I1 acid per gram alumina B = Milligrams Orange I1 acid per cc solution A
66 7’0 1163 1808
B 0.640 0.640 0.615 0.640
A
B
2150
2890 3340
0.625 0.625 0.615
3980
0.655
The entire range gives a practically constant value for the concentrations of the solutions, thus showing the existence of two phases and confirming the conclusion of Gordon that a definite chemical compound is formed under these conditions. Qualitative observation of the systems during the progress of the reaction bearsout this conclusion. The Orange I1 lake settles out immediately on mixing and is probably then an adsorption complex. After standing for five days at room temperature needle-like crystals could be observed in all the flasks. Some of these crystals were over an inch long. Careful inspection showed that the powdery mass of almina had been replaced by a mat of crystals. Analyses of some recrystallized samples gave 96.7 percent dye acid instead of 97.3 percent required by AlX8. That is close enough for our purposes. One sample of an aqueous alumina suspension, which was about one year old, was treated with an equal volume of the dye acid. Crystals of the aluminum salt were noticed in a few hours. A sample of alumina prepared by heating aluminum nitrate was treated in the same way. This alumina had been shown previously to have practically no adsorbing power for alizarinic acid. With Orange I1 acid a slight swelling of the particles was noticeable after forty-eight hours. After standing for two weeks the appearance of the sample was very interesting. The large particles of alumina retained their original form; but small radiating crops of needles seemed to sprout from them. At the end of two months the alumina had disappeared partly and a good deal of crystalline material had formed. When alumina was treated with aqueous solutions of the sodium salt of Orange 11, the result was apparently an adsorption isotherm as shown in Table I11 and Fig. 2. 1
Knecht and Hibbert: “New Reduction Methods in Volumetric Analysis” (1925).
3130
WILDER D. BAXCROFT AXD ESTHER COE FARKHAM
TABLE I11 Alumina and Sodium Orange I1 in Water Length of run: three weeks A = Milligrams Orange I1 adsorbed per gram alumina B = Milligrams Orange I1 per cc solution A 411
B 1.70
A
B
1155
865
3.32
I490
5.25 7.02
FIG 2 Sodium Salt of Orange I1 in Water and Alumina
This unexpected result might be due to one of two causes: Orange I1 is adsorbed in the form of the sodium salt (the obvious I. explanation). The dye reacts with alumina to form a definite compound; but 2. the resulting change in caustic soda content is sufficient to cause a variation in the solubility of the compound, which simulates adsorption. The second alternative is ruled out because the pH of the solutions varied between 7 . 2 and 7 . 6 , whereas, it should have been between 11.0 and 1 1 . 8 if all the sodium hydroxide corresponding to the dye had remained in the solution. On the other hand special analyses showed no decrease in the amount of sodium in the solution. This is analogous to what was found by Weiser for sodium alizarate; the explanation is undoubtedly the same in the two cases.
3131
ALUMINA LAKES
Our alumina was impure and contained ammonium chloride. The Orange I1 anions are exchanged for chloride ions and the sodium ions left in solution. If we were to start with a pure alumina, made from amalgamated aluminum, we should undoubtedly get adsorption of the salt of Orange I1 acid just as was done in the case of sodium alizarate. Unfortunately, time did not permit of this experiment being done at present. I n order to tie in the results with Orange I1 acid and those with alizarinic acid, a set of experiments was made with alcoholic Orange I1 acid and an alcoholic suspension of alumina. The results are given in Table IV and Fig. 3.
FIG.3 Orange I1 Acid in Alcohol and Alumina
TABLE IV Alumina and Orange I1 Acid in Alcohol Length of run: three weeks A = Milligrams Orange I1 per gram alumina B = Milligrams Orange I1 per cc solution A
B
A
498 1195 2425
0.046 0.197 0.334
3810 5030
B 372 378
The curve shows a first portion which seems to indicate adsorption, followed by a flat due to the formation of the chemical compound obtained by Gordon and confirmed by us. It was noticed that crystals of the aluminum salt were plainly visible in the last two reactions flask, while no crystals could be detected in the first three. The alumina, however, was colored orange. Since the aluminum salt of Orange I1 acid is appreciably soluble in alcohol, the first portion of the curve is of an adsorption curve modified by solubility. For extreme accuracy a correction would have to be made for the amount of aluminum salt in solution. This correction has not been made. I t is not clear why Orange I1 acid should form a compound readily with alumina while alizarinic acid does not. Orange I1 is a stronger acid than alizarinic acid; but that is not the whole thing because sulphuric acid, which is a still stronger acid, is adsorbed by alumina from low concentrations in aqueous alcohol.
3132
WILDER D. BANCROFT AND ESTHER COE FARNHAM
We hoped to be able to show that alumina precipitated from a sulphate solution would take up less alizarin than alumina precipitated from a chloride solution, because of the greater adsorption of sulphate over chloride. It is well known that some acid dyes which are not affected by sodium chloride can be stripped from wool to some extent by sodium sulphate. This result was not attained because, under the conditions of our experiments, alizarin ions are adsorbed so much more strongly than either sulphate or chloride ions that any difference between the last two is negligible. By adopting a new method of precipitating lakes we were able to get some new and interesting information in regard t o alumina-alizarin lakes. The standard method of making alumina-alizarin lakes is: t o precipitate the alumina with ammonia, sodium carbonate, or caustic soda; to wash it several times by decantation or by centrifuging; and then to add the dye to the alumina suspension, usually following this with addition of a solution of calcium acetate to effect more complete precipitation of the dye. Since we wished to study the effect of various salt ions, we substituted for this procedure one which we call the one-step process. This consists in treating a known solution of an aluminum salt with the calculated amount of standard alkali, adding the desired amount of dye immediately. A later improvement by Dr. H. L. Davis is to mix the dye and the alkali, adding this mixture to the solution of the aluminum salt. The advantages of the one-step method for our purpose are three-fold: I t saves the time and loss due to washing. I. I t makes possible the study of the effect of the salts present (KaCl, 2. Na2S04, etc.). 3. It eliminates practically completely effects due to the ageing of the undyed alumina. I t became apparent a t once that, in order to use this method to advantage, we needed to know more about the efficiency of alkali and of dye salts in precipitating alumina from solution in the form of a lake. This problem has been discussed in part in the paper,’ on “Titration Curves for Aluminum Salts with Alkalies.” The results which are now given have to do with the effect of sodium alizarate as a precipitating and dispersing agent.* The solutions used were normal aluminum chloride and aluminum sulphate, standard 0.25 K sodium hydroxide and 0.024 N (0.012 M) sodium alizarate. Table V shows a preliminary run with constant equivalent amounts of aluminum chloride and aluminum sulphate, and varying amounts of dye and alkali. The solutions were mixed in the order previously described, the total volume was made up to 20 cc, and observations were made at the end of twenty-four hours. The data are given in Table T’. Davis and Farnham: J. Phys. Chem., 36, 1057 (1932). This property of alimrin has been known for some time. Knecht (“Manual of Dyeing,” 2, 574) states that “an excess of alizarin prevents the precipitation of aluminum hydroxide from its salts with ammonia.” 2
ALUMINA LAKES
3133
TABLE V Precipitation of Aluminum Salts by Sodium Alizarate and Sodium Hydroxide One aluminum atom is equivalent to three mols NaOH A
z cc N AlCl3
Equiv. cc Observations NaOH dye 2 z no precipitate formed
complete precipitation and exhaustion partial precipitatioq; supernatant liquid light orange 4 slight precipitation; no exhaustion 4 nearly complete precipitation; supernatant liquid slightly yellow 4 partial precipitation; exhaustion not complete; supernatant liquid light red B z cc N Alz(S04)3 2
3
2
4 2
3 4
nearly complete precipitation; supernatant liquid light orange complete precipitation and exhaustion partial precipitation; dark red lake; supernatant liquid light orange 2 4 practically complete precipitation; supernatant liquid light orange 4 complete precipitation and exhaustion 4 partial precipitation; supernatant liquid light orange 2
2
3
2
4 z
3 4
An examination of Table V shows that the precipitation range for the sulphate lakes is wider than that for the chloride lakes, which is in accord with what had been found previously for the action of alkali alone upon aluminum salts. Sodium alizarate behaves in general something like sodium hydroxide, causing precipitation in small amounts and dispersion in large amounts. I n Table VI are given results when the aluminum chloride or sulphate was kept constant a t 2 cc, and the caustic soda at twoequivalents, whiletheamount of dye varied. I n order to simplify the tabulation, the following symbols are used for describing the tubes: Indicates a clear solution; red with no precipitate. Indicates partial precipitation of alumina. a) complete exhaustion. b) incomplete exhaustion. (degree of exhaustion is indicated by minus signs for very slight color in the supernatant liquid and by plus signs for much color). 3 . Indicates complete precipitation of alumina. a) complete exhaustion. b) incomplete exhaustion. I t is somewhat difficult to judge between 2 and 3. The variation in the volume of the precipitate was used as a preliminary indication and confirmed by testing the liquid with ammonia. I.
2.
WILDER D. BANCROFT AND ESTHER C O E FARNHAM
3134
TABLE VI Precipitation and Peptization of Alumina Lakes by Varying Concentrations of Sodium Alizarate A = 2 cc N AIC13; B = 2 cc K Alz(S04)Z Equivalents of sodium hydroxide = 2 0 = observations after twenty-four hours ccdye
0-A
0-B
I
2b 3a zb
za 3a 2b
I
2
3 3 5
6
ccdye
0-A
0-B
7 8
I
I
I
I
9
I
I
I
IO
I
I
I
I
I1
I
I
I
I
I2
I
I
+
At the end of twenty-four hours all those tubes which did not give clear, colorless, supernatant solutions (those not marked 2a or ja) were treated as follows : Five cc of the supernatant liquid were drawn off and placed in a small testtube. To the aluminum chloride solutions there were added 5 cc of saturated sodium chloride solution, while an equivalent amount of saturated sodium sulphate solution was added to the aluminum sulphate solutions. The tubes were stoppered, shaken, and examined after having stood for twenty-four hours. This use of two different salt solutions is not the usual procedure. I n these experiments the object was to have the two systems exactly comparable except for one variable, the inorganic anion. By using the same anion in the aluminum salt and in the added salt it is possible to observe the difference due t o only one cause with no complicating factors to consider, such as would result if we used chloride and sulphate initially and sulphate finally in both cases. This would give in one case three competing anions, sulphate, hydroxyl, and alizarate; in the other case four, sulphate, hydroxyl, alizarate, and chloride. The results are given in Table VII.
TABLE VI1 Flocculation of Peptized Alumina-Alizarin Lakes A = aluminum chloride lakes with sodium chloride B = aluminum sulphate lakes with sodium sulphate 0 = observations after twenty-four hours cc dye = concentration of dye per 2 0 cc in original 0-A 38
0-B
ccdye
I 2
-
-
3% 3a 3a 3b---
3a 3a 3b--3b---
7 8 9
ccdye
3 4
5 6
-
IO II
12
0-A
*
0-B I
)F
I
3b 3b 2b(:)
3b 2b+ 2b++
I
* These tubes show the full red color; but a partial precipitation of the lake can be de-
tected by a cloudiness a t the bottom of the tubes.
ALUMINA LAKES
3’35
All tubes containing more than 3 cc dye fail to form precipitated lakes. This is due to the fact that the dye anions, being adsorbed strongly, exert a peptizing action upon the lake which is comparable with that of excess sodium hydroxide, For alumina alone about 4.2 equivalents of alkali are sufficient to redisperse alumina in an aluminum chloride system, while 4.7 equivalents is needed for the sulphate system. If we consider 4 cc of dyes being the first completely redispersed system in the presence of two equivalents of caustic soda, we must then assume that 4 cc of the dye are equivalent to a t least 2 . 2 NaOH. Since the dye solution is 0.024 normal, 4 cc = 0.0096 normal cc. 2 equivalents of NaOH = 1.33 normal cc per 2 cc AlC18. 4.2 equivalents of 0.096 = 1.429 cc. It is thus evident that the dye NaOH = 1.4 cc and 1.33 solution is about equivalent in peptizing power to the same amount of sodium hydroxide. This knowledge is very useful in the preparation of alizarin lakes by the one-step process. I n the chloride suspensions of alumina a positively charged lake is formed in the presence of I cc of dye. This is due to the fact that an excess of aluminum chloride is present, the peptizing power of the strongly adsorbed, trivalent aluminum ion overcoming the effect of the weakly adsorbed chlorine ions and stabilizing the ~01. In the sulphate solutions the bivalent sulphate ions are adsorbed strongly enough on the acid side to prevent this action. The addition of sodium chloride or sodium sulphate flocculates the sols resulting from the presence of an excess of dye, even though these are negatively charged sols. The flocculation must be due to an adsorption of sodium ions in preference to chlorine and sulphate ions. With the negatively charged sols there is thus a reversal of the anion effect, sodium sulphate having less flocculating power than sodium chloride because sulphate ion is adsorbed more strongly than-chloride ion and therefore tends to keep the sol more negative. This is identical with what was observed years ago with albumin sols. Both alumina sols show a region of maximum stability on addition of a definite amount of sodium chloride or sulphate. For this concentration this region occurs with 7-9 cc dye. The sulphate sols remain perfectly clear for twentyfour hours after the addition of salt, while the chlorides are precipitated partially. Beyond this range is another of comparatively low stability, followed by a third, stable, repeptization zone. This is a striking example of the wellknown irregular series.’ It is easy to see that the action of alkali on aluminum sulphate and aluminum chloride should produce a precipitate at lower concentrations in the sulphate solutions; but it is not a t all clear why more alkali is necessary to peptize the alumina from the sulphate solution. It should be the other way round. This probable explanation is that the difference is apparent and not real, being due to the fact that the alumina from the sulphate solution is much denser and is apparently less hydrous than that from the chloride solution. Bancroft and Ackerman* have accounted for the polygenetic nature of alizarin with different mordants by postulating the presence of yellow, un-
+
2
Kruyt and van Klooster: “Colloids” (1927). J. Phys. Chem., 35, 2568 (1931).
3136
WILDER D. BANCROFT AND ESTHER COE FARNHAM
dissociated alizarinic acid, red alizarate ion, and purple undissociated sodium alizarate. Dr. H. L. Davis has pointed out to us that ii would be more in line with the views of Hantzsch’ to postulate the existence of several, colored, tautomeric, alizarinic acids. Sorensen* describes the use of alizarin as an indicator. He records two distinct color changes: yellow to red at pH 5-5-6.8, and violet to purple at pH 1 0 . 1 - 1 2 . 1 ; the change from red to violet is gradual. By very careful addition of alkali to a dilute solut,ion of alizarinic acid, one easily obtains all four colors: yellow, red, violet and purple (blue). Even assuming that the violet form is a mixture of red and the blue, there are three, distinct, colored forms of alizarinic acid, and it is necessary to assume some taut,omeric changes to account for these differences. The nature of the solvent often has a great effect on the equilibrium between two forms, but it should have no effect on the nature of the form crystallizing from the solution, provided the crystals do not contain solvent of crystallization. Meyer and J a c o b ~ o nnot , ~ knowing the phase rule, state that alizarinic acid crystallizes in yellow crystals from alcohol, but in orange-red crystals from other organic solvents. Special experiments showed that solutions of alizarinic acid in chloroform, carbon tetrachloride, ether, n-butyl bromide, and paraldehyde are yellow, solutions in alcohol and acetone are orange-yellow, and solutions in benzaldehyde are orange. Alizarin crystallized from these solvents in yellow to yellow-orange crystals, depending on the size of the crystals. When powdered, all the crystals were the same yellow. The behavior of a pyridine solution is interesting. In pure, dry pyridine alizarin dissolves to form a yellow solution. On dilution with a small amount of water the solution becomes red,changing to purple as still more water is added. This change is in the direction one would expect from increasing alkalinity. If the polygenetic character of alizarin is due to tautomerism and if this tautomerism is affected by the pH of the solution, it should be possible to change the color of alizarin lakes by varying the pW. Experiments were therefore made with alumina and tin lakes. I t was found possible by special manipulations to change the color of the alumina-alizarin lake through the entire series from yellow to violet and back t o red. The alumina was prep:>red by heating aluminum nitrate. While this alumina does not take up much alizarin, it has the advantage of not being peptized readily by acid or alkali. A dilute solution of alizarinic acid in alixarate alcohol was t,reat,edwith a slight excess of caustic soda. ~~‘hilesodium is not very soluble in alcohol, a purple, apparently clear, solution can be obtained. When shaken with alumina a lavender lake was obtained, which changed to pink on washing with water. Washing with a three percent tartaric acid solution changed t,he color to yellow. Washing with caustic soda solution restored the original lavender color. A red lake was prepared by treating alumina wit,h alcoholic alizarinic acid. When treated wit,h a solution of phosphoric acid, the lake became yellow. Washing with caustic soda solution changed the lake to a violet red. 1
2
3
Cf. Heinrich (Johnson and Hahn): “Theories of Organic Chemistry,” 382-478 ( 1 9 2 2 ) . Biochem. J., 21. 241 (1909). “Lebrbuch der organischen Chemie,” 2, 11, 555 (1905).
ALUMINA LAKES
3'37
Tin mordant was prepared by the method described by Ackerman. I t was suspended in alcohol and treated with an alcoholic solution of alizarinic acid. A yellow lake was formed. The mixture was made alkaline with ammonia, after which acetic acid was added drop by drop until a red shade resulted. On filtering, a red lake was formed which was fairly deep in color and reasonably fast to washing with alcohol and water. The color is actually more nearly an orange red than the cherry red of the alumina lakes. With sodium alizarate and alkali tin gives a purple lake. We see that alumina lakes are red over a wide range of pH; but can be obtained yellow or purple. With stannic oxide the red lake exists only over a narrow range of pH, the yellow occurring over most of the acid range and the purple over most of the alkali range. With other mordants one would probably get ranges intermediate between these two extremes. The general results of this investigation are: I . By using alcohol as a solvent it is possible to make a phase rule study of alizarinic acid and alumina. 2. Kahlbaum's alumina is practically inert towards alizarin. Alumina obtained by heating aluminum nitrate adsorbs alizarin slightly. A precipitated alumina dried at 100' is a moderately good adsorbent. Freshly precipitated, washed alumina is a good adsorbent. 3. At ordinary temperatures alizarin forms no definite chemical compound with alumina. The alumina-alizarin lakes are adsorption complexes. This confirms the results of Weiser and Porter. 4. Orange I1 acid forms a definite chemicalcompound, AlX3,withalumina both in aqueous and in alcoholic solution. This confirms the results of keinmuth and Gordon. At low concentrations Orange I1 mid is adsorbed by alumina. This was not discovered by Reinmuth and Gordon, who did not study this range. 5 . The sodium salt of Orange I1 is adsorbed by alumina, no definite chemical compound being formed. This was not discovered by Reinmuth and Gordon. 6. Caustic soda and sodium alizarate have been shown to be nearly equivalent in the precipitation and dispersion of alumina. ;. The negatively charged lakes formed by peptization with sodium alizarate are destabilized by sodium chloride and sodium sulphate owing to adsorption of sodium ions. There is a reversal of the order of the anions just as there is with albumin. 8. The color of an alizarin solution can be varied over the entire range from yellow to red to violet (to blue) by a progressive increase in pH. Since this is apparently due to tautomeric changes, one must at least assume the existence of yellow, red, and blue (purple) alizarinic acids. 9. The colors of aliearinic acid dissolved in pyridine can be varied through the entire range by adding water which increases the alkalinity. IO. Under suitable conditions one can prepare alumina and tin lakes of all the colors. With alumina the red is stable over a wide range of pH and with tin over a narrow range. Cornell University.
